SearchMash: Google`s New Experimental Playground

When you operate a search engine that is queried nearly a billion times a day by hundreds of millions of users around the world, there are certain things you can’t do. For example, you can’t just make the kinds of wholesale changes and experiments that gave you the technological edge to make you the leading search engine in the first place. So what’s a market leader to do?

Well, when your name is Google, you answer this quandary by building a whole new search engine. Meet SearchMash (http://www.searchmash.com), a web site owned by Google but too shy to say so unless you check out its privacy policy. While the interface on the home page is every bit as clean as Google’s, there’s nary a Google logo in sight:

 

A good-sized white box is surrounded by a periwinkle-colored backdrop, and contains links labeled Features, About, Privacy, and Terms of Service. The copyright is dated 2006; it mentions SearchMash and only SearchMash.

The obvious question for most Google mavens, before we even start performing searches and clicking links, is why is SearchMash a freestanding website and not in Google Labs? Danny Sullivan asks Google exactly that. He was told that “one of the important factors we wanted to address was the influence that may come from Google branding. Creating a separate site will help us gather more objective data about user response to new interfaces.”

Translation: too many people automatically think that anything coming from Google is cool and relevant. We want to get away from that. It may sound a little arrogant, but there are actual studies that back this up. In late April, Barry Schwartz reported on some interesting findings discussed at the SES Toronto’s Searcher Behavior Research Update panel. Gord Hotchkiss, one of the panelists, said that one of his eye tracking studies revealed that Google users were much less likely to look past the second or third result returned than those who used MSN or Yahoo. Lance Jones, another speaker on the panel, reported that one of his studies showed that users rated search results they knew came from Google as more relevant – even when the two sets of results they rated were identical.

Given SearchMash’s position as an experimental playground, you can expect its feature set to change; the ones I discuss in this article are the ones that were available for me to try out at the time of this writing. Oh, and that’s another point; Google said of SearchMash that “This site is only a test and has traffic limitations so may be unavailable at times.”

Click on the Features link, and you get a list of current features that explain them in both words and screen shots. Next to each feature is the question “Is this useful?” and you can click on “yes” or “no.” The ability to vote makes a lot of sense. Google specifically said that “There is no guarantee that the features tested on SearchMash will be seen on Google Search. As with all of our experiments, one of the main factors we will consider is user response to the feature and how well it addresses their needs.”

The “start typing” feature was nothing special. It automatically lets you start typing in the search box without your cursor needing to be there. It’s not actually limiting where your cursor can go; if you click somewhere else on the page, then start typing, what you type doesn’t show up in the search box. It might save a little time by letting you type your queries as soon as you arrive on the page.

The URL menu, on the other hand, is kind of nifty.

 

By clicking on the green URL that appears below each result, you can choose a number of options pertaining to that result. You can display it in the same window, a different window (handy if you like to refer back to your search results), or open a cached copy (which brings up a Google-branded cached copy of the page). “More from this site” turns up a search result listing more pages from the same site (nearly 3,000 pages in the case of SEO Chat), while “More similar pages” turns up similar pages regardless of web site.

That last feature needs work. The top result was highly relevant for the search I performed, but others seemed to be all over the place. A number of search engines turned up, which made sense if you thought in terms of “related” rather than “similar” (indeed, clicking on the link for “More similar pages” changed the search in the search box to “related:www.seochat.com/”). There’s a difference in the shades of meaning between “related” and “similar” that I guess is tricky to duplicate in an algorithm.

The image results panel is a nice touch. It shows the top three images for your search next to the web results, on the right hand side. I don’t think of search engine optimization as particularly photogenic, but here is the panel SearchMash gave me when I did a search for SEO:

 

As you can see, it tells you how many images turn up for your search term, and gives you the option of clicking to see more images. When you click on the image, it takes you to the web site on which the image appears, which might or might not be related to the image – or useful to your search. The cartoon in the center adorned a story about a strategic partnership to provide a suite of press release SEO tools (which is at least vaguely related). The intriguing flow chart on the bottom led to a site in Japanese – fortunately for this poor American, the site offered an English translation.

The more images link takes you, naturally enough, to a page with more images. With more than 200,000 hits for matching images, you would expect to see links to numbered pages at the bottom. Instead, you get a page that barely scrolls vertically and another link at the bottom that says the same thing, more images. What gives?

That can best be answered by explaining another feature of SearchMash: the more results bar, which is also used here. At the bottom of your search results, SearchMash gives you a link that says “more web pages.” Click on that link, and the page automatically scrolls upward, giving you another 10 web results. The same thing happens with the more images link: click it, and the page scrolls up, giving you about another screen’s worth of images from which to choose. This feature is very similar to the “continuous scroll” once used by Windows Live and discontinued for web site results (though Amazon added it to A9 after a recent revamp that threw out many of its features).

Finally, I’d like to touch on a feature SearchMash says is “just for fun right now, but we have some ideas for how to use this.” It’s called reorder results. You can click the number next to a search result and drag it around to change the order. The number stays the same though. You can see that in this screen shot of a reordering I did for a search on juggling equipment.

 

You’re probably wondering why I did juggling equipment instead of SEO. Actually, I did do an SEO reordering the day before I did this one. SEO Chat came up in the number three position, so just for fun I moved it to the number one spot. When I came back to finish this article the next day, SEO Chat came up in the number one position rather than the number three spot. (It looks like we’re the second organic result on Google for SEO at the time of this writing, so that doesn’t necessarily prove anything).

SearchMash is Google’s testing ground for search engine technology. As such, it’s unlikely that we’ll ever see any ads on this site, unless there’s something about ad technology that Google wants to test – and even then, if I were Google, I might use a different site. (Google is certainly able to do that; the company has bought up a ton of variations on the SearchMash URL).

A number of observers have commented that the reordering search results feature might be a way for Google to let users “tweak” the results so that they are more personally relevant. It could also be a way for users to “vote” on the relevance of particular results and thus influence a web site’s actual standing in the SERPs. If that is the case, I’m sure I don’t need to mention how open this could be to abuse. It’s not a feature implemented in Google itself, of course, and I imagine the search engine would think long and hard before adding it.

Danny Sullivan has pointed out that, with SearchMash being covered in blogs and the SEO-centric press, it is attracting exactly the kind of test audience that Google doesn’t want: “people who are the influencers or tech-heads or early adopters that Google’s not trying to test against.” Google apparently has ways to filter out that kind of traffic and drive the other kind of traffic to SearchMash for particular test purposes. At the very least, Google can put SearchMash in front of focus groups for unbiased opinions of features without even letting them know the true source of the technology.

Speaking of technology, it shouldn’t surprise anyone that many of SearchMash’s features seem to be powered by AJAX, since Google has already shown signs of heading in that direction. To mention another insight from Danny Sullivan, SearchMash feels somewhat similar to Amazon’s A9 – which is interesting in light of A9′s CEO Udi Manber defecting to Google.

What does this mean for SEOs? If you want to see what Google is working on, you can go to SearchMash. But my guess is that these are longer-term projects – some engineer’s “20 percent time,” as Lee Odden of Searchnewz speculated. I’m quite prepared to eat my words if I’m wrong, but I don’t think we’ll see any features migrating from SearchMash to Google for at least six months to a year (and probably longer). Still, watching Google’s playground to see what it is up to ought to be fun. I wonder how long it will be before SearchMash includes a form for users to comment on what features they’d like to see next.

Google+ Comments

Google+ Comments